The criteria, in the order they matter
1. How big is the load? Heating and cooling is almost half the bill by ENERGY STAR’s own figure. A lamp is tens of watts. A charger left in the wall is a fraction of a watt. If a device does not touch the heating, the cooling, or something that runs for hours a day, the most it can save you is small — however clever it is.
2. Does it measure, or only switch? You cannot cut what you cannot see. Two of the plugs we carry publish energy monitoring on the maker’s own page; two do not. A plug with no meter can still save something by turning a load off on a schedule, but it will never tell you which load was worth attacking.
3. Does the saving survive the install? The cheapest thermostat is worthless if your wall has no C wire and the box has no adapter. This is the most common reason a smart thermostat never gets fitted, and it is a published, checkable fact per model rather than a matter of opinion.
4. Is there a fee? None of the five picks on this page requires a subscription for its core function. That is the house rule on this site, and it rules out more products than you would expect.
5. Does it still work when the internet is out? A schedule that stops running when your broadband drops is a schedule you cannot rely on. TP-Link publishes local control for its plugs; most other makers say nothing either way.
Why the ecobee leads and the Nest sits one place behind
These two are close, and which is right for you turns on one question we come back to below. The ecobee takes the top slot here for three published reasons, each of which attacks the biggest load in the house.
It is ENERGY STAR certified, which ecobee publishes across the line, and that certification is earned on field data rather than a lab claim. It ships a SmartSensor in the box, which matters more than it sounds: a thermostat with one sensor regulates the hallway it is screwed to, and a setback that keeps the hallway comfortable while the occupied room drifts is a setback people cancel within a week. And it ships a Power Extender Kit, so the most common install blocker — no C wire in an older house — does not stop the project. ecobee also publishes a savings claim of “up to 26%*”, footnoted as measured against a hold of 72°F in its own April 2021 analysis. We report that as ecobee frames it rather than endorsing it: a comparison against a constant 72°F hold is a comparison against one of the least efficient ways to run a house.
Honest cons: it is the priciest thermostat we list, it is physically larger than the alternatives, and some extras vary across the Premium / Enhanced / Essential line, so check which model a listing is actually selling before you buy.
The Nest Learning Thermostat (4th gen) is second because its saving mechanism is behavioral rather than certified. Google publishes that the thermostat builds a basic schedule for you and that Home/Away “learns when you are home or away to save energy”. And if your heating is a heat pump, Google publishes the one figure on this page that genuinely changes a purchase: “AUX heat can cost about 2 to 5 times as much as running your heat pump”, with a Heat Pump Balance feature that sets the aux lockout temperature automatically from how your heat pump is performing. On a heat pump that single behavior outweighs everything else on this page — which is why our heat pump thermostat ranking puts the Nest first and this page does not. Different criterion, different winner, and we would rather say so than pretend one order fits every house.
Honest cons on the Nest: it may still need a C wire or Google’s separately sold power connector, Google’s own help pages document real limits on Alexa control, and we did not find a first-party ENERGY STAR statement for it on the Google pages we read for this guide — which is not a claim it lacks certification, only that we will not assert one we did not read.
Measure before you optimize
The second half of this page is the part most guides skip. Before you buy anything to cut usage, it is worth knowing where the usage is. Two of the plugs we carry publish energy monitoring on TP-Link’s own product pages, and they differ enough to matter.
The Tapo P110M publishes real-time and historical energy statistics with bill estimation in the app, and it is Matter certified, so the same plug joins Apple Home, Alexa, Google Home or SmartThings without a brand bridge. Two of its features are about waste rather than convenience: a Smart Charge Guard that cuts power once draw falls below a threshold you set — the fix for a laptop or a battery that sits at 100% trickling all night — and a Power Protection cut-off. TP-Link also publishes local control: “Your connected devices in your local area network (LAN) remain accessible via the Tapo app, even when the household goes offline”.
The Kasa Smart Plug Mini (EP25) is the cheap route to the same visibility. TP-Link publishes real-time and historical energy monitoring for it in the Kasa app, it comes in a four-pack, and unusually at this price the page lists Siri alongside Alexa, Google Assistant and SmartThings. What it does not claim anywhere on TP-Link’s page is Matter certification, so it reaches Apple Home through TP-Link’s own HomeKit support rather than through Matter.
One limit applies to both and is worth reading before you buy: TP-Link publishes a maximum of 15A / 1800W resistive for the P110M (1/6 HP motor), and 15A / 1.8kW with the same motor rating for the EP25. A plug-in space heater sits right at that line, and “smart” is not a reason to run one harder. If the thing you want to meter heats or has a motor, check its rating first.
Where a smart plug saves you nothing
A plug does not make the appliance behind it more efficient. It saves energy in exactly two situations: when it stops something running that nobody is using, and when it tells you something is drawing more than you assumed. For a fridge, a router, a freezer or anything else that must run continuously, a smart plug saves nothing at all and adds a failure point. Our do smart plugs save energy guide works through the arithmetic, including the cases where the honest answer is no.
The same caution applies to standby draw. We would like to tell you what share of a U.S. household’s bill goes to devices sitting idle, but we could not reach a first-party federal page publishing a current figure while writing this, and secondhand numbers circulating from old studies are not something we will print as fact. What we can cite is a manufacturer figure: TP-Link publishes standby power for the Kasa KL135 bulb as “as low as 0.2 watts”. That is the order of magnitude per device — small individually, and worth metering before you buy hardware to fix it.
Lighting: watts are the easy part, hours are the real one
Replacing incandescent bulbs with LEDs is the oldest advice in home energy, and ENERGY STAR still publishes the reason: LED lighting uses a fraction of the energy of incandescent for the same light output, and is rated on lifetime rather than burnout. That part of the saving happens whether or not the bulb is smart.
What a smart bulb adds is control of the hours. The Kasa KL135 publishes 1,000 lumens with a dimming range of 20lm to 1,000lm at a 60W equivalent, brightness from 1% to 100% in the app, grouping, and voice control with Alexa or Google Assistant. Dimming to 50% is not a 50% saving, but a light that is off when the room is empty is a 100% one, and a bulb you can group and schedule is a light that actually gets switched off.
Honest cons: TP-Link publishes no Kelvin range, no rated lifetime hours, no Wi-Fi band and no wake-up dimming feature for this bulb, so we describe none of those. It is also a single bulb — if you are lighting a whole room, our smart light bulb ranking covers the bridge-versus-Wi-Fi decision that arrives with scale.
The five picks, scored on what each one actually controls
'Not published' means we read the maker's own page on the date in the sources below and found no statement; 'not read for this page' means we did not verify that specific claim for this guide and will not assert it either way. Neither is a claim the feature is absent. Savings depend on your climate, your home and your tariff — nothing here is an estimate for your house. | ecobee Premium | Nest Learning (4th gen) | Tapo P110M | Kasa EP25 | Kasa KL135 |
|---|
| Load it controls | Whole-home heating + cooling | Whole-home heating + cooling | One appliance per plug | One appliance per plug | One light fitting |
|---|
| Publishes energy monitoring | Not read for this page | Not read for this page | Yes — real-time and historical, with bill estimation | Yes — real-time and historical | Standby draw only (0.2W) |
|---|
| ENERGY STAR certified | Yes — published across the line | Not read for this page | Not applicable | Not applicable | Not published |
|---|
| Likely install blocker | C wire — Power Extender Kit included | C wire — may need Google's power connector, sold separately | None — plugs in | None — plugs in | None — screws in |
|---|
| Works without a subscription | Yes | Yes | Yes | Yes | Yes |
|---|
| Keeps working if the internet drops | Not read for this page | Not read for this page | Yes — TP-Link publishes LAN control | Not published | Not published |
|---|
| Ecosystems | Apple Home, Alexa, Google, SmartThings, Matter | Google Home, Alexa, Matter, Thread | Apple Home, Alexa, Google, SmartThings, Matter | Apple Home (Siri), Alexa, Google, SmartThings | Alexa, Google |
|---|
| Published maximum load | Not applicable | Not applicable | 15A / 1800W resistive, 1/6 HP motor | 15A / 1.8kW, 1/6 HP motor | Not applicable |
|---|
What we left out, and exactly why
The Kasa EP40M outdoor plug. It is the right plug for holiday lights and a porch lamp, and it is Matter certified with two independently schedulable outlets — but TP-Link publishes no energy monitoring for it, so on a page ranked on measurement it does not earn a slot. It is a pick on our smart plug ranking instead.
The Wyze Plug. Wyze publishes no energy monitoring, no Matter and no Apple Home support for it. It is a fine switch for a house already on Wyze; it cannot tell you anything about consumption.
A whole-home panel monitor. A clamp-on monitor in the breaker panel is the correct answer to “where is my electricity actually going?”, and we do not carry one. We checked the live catalogue while writing this page: the obvious candidate returned three separate listings at three different prices under one identical product title, with no published way to tell which sensor count each listing ships. We will not guess which box a reader would receive, so there is no pick here rather than an unverified one. Worth knowing either way: a panel monitor measures rather than saves. It tells you where to act, and you still have to act.
Anything that charges a monthly fee to report your own consumption back to you. None of the picks above does.
A sensible order to buy in
If you are starting from nothing: fit the thermostat first, because it is the only device here with access to half the bill. Live with it for a season before you buy anything else — a learned schedule takes time to be worth judging. Then add one metering plug, not four, and move it around: the dehumidifier, the second fridge in the garage, the games console, the towel rail. The job of the first plug is to find out which loads deserve the other three.
Lighting comes last, not because LEDs are not worth it — they are, and ENERGY STAR explains why — but because the saving per fitting is small enough that it should not stand between you and a thermostat. If you want the whole-house picture first, our smart home running cost guide covers what the devices themselves draw, and do smart thermostats save money separates what is attributable to the hardware from what is just a schedule you could have set by hand.
